CONTROL AND COMPUTATION FOR MECHANICAL MANIPULATORS
Abstract
The main area in the research includes parallel computation of input torques, straight-line segments path planning, and approximation of joint trajectories. The advantage of using a multiprocessor controller for a mechanical manipulator is that parallel computations can be arranged to achieve a minimum computing time. To schedule tasks for parallel processing, first the entire task is divided into subtasks. Based on the precedence relations, an optimum order to execution for each CPU is obtained by using a scheduling algorithm, which is developed base on an idea extended from the method of branch-and-bound. The path planning deals with the motion of the hand along straight line path. While considering constraints on the maximum acceleration and velocity of the hand, a large number of nonlinear inequality constraints are involved. Method of Approximate Programming (MAP) is adopted and modified to obtain a minimum traveling time. Methods with concepts of spline functions and least square error fit are used to approximately express a Cartesian path of a hand in joint trajectories. Joint position, velocity and acceleration are all kept continuous at any point where two polynomials are splined together. Such approximation allows one to control the traveling path of a hand, which is specified in the Cartesian coordinates, at the joint level, and thereby reduces some computations for the coordinates transformation of the path.
Degree
Ph.D.
Subject Area
Electrical engineering|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.